Understand this story
On November 21, 2026, the SimpleBench team launched SimpleBench, a multiple-choice text benchmark for large language models. The evaluation contains more than 200 questions covering spatiotemporal reasoning, social intelligence, and robustness against linguistic adversarial attacks. Benchmark results show that ordinary people with high-school-level general knowledge scored 83.7%, outperforming all tested large language models; Claude Fable scored 81.9%. The human score was based on a small sample of nine participants. There are currently no further developments supported by adequate factual evidence.
AI-generated from public reports
Latest development · Report
SimpleBench: Humans Outperform All Tested LLMs on a Basic Reasoning BenchmarkDevelopments
1 confirmed developments
SimpleBench Team Launches Multiple-Choice Text Benchmark for Large Language Models
Source:English text is being prepared.
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 21
Trending on Hacker News
SimpleBench: Humans Outperform All Tested LLMs on a Basic Reasoning Benchmark
SimpleBench shows that ordinary people with high-school-level general knowledge outperformed all tested LLMs, scoring 83.7% on the benchmark, compared with Claude Fable's 81.9%. The multiple-choice text benchmark contains more than 200 questions covering spatiotemporal reasoning, social intelligence, and adversarial robustness in language; the human score is based on a small sample of nine participants.
AI Trends